During tumorigenesis, cancer cells as well as healthy cells depend upon growth factors produced in an autocrine and paracrine manner to maintain their proliferafion and differenfiafion. Hence there exists a "tug-ofwar" for growth factors between a tumor and the surrounding non-cancer cells, and the fitness of each individual cell must be defined by this competifion at the populafion level. Importantly, cancer cells on the leading edge of tumors are characterized as being chemo-resistant, have enhanced metastatic potential and are extremely efficient at forming new tumors [6]. We have determined that cancer cells and adjacent stromal cells express high levels of the growth factor IL-6 and activated Stat3 which are hypothesized to be the principal mediators of both tumorigenesis and metastafic progression. In this proposal, we focus on the IL-6/pStat3 pathway in melanoma and breast cancer cells, whose proliferation and differenfiafion crifically depends on pStat3 signals. In the context of understanding tumor dynamics, especially during drug treatment, the intraclonal competifion for growth factors within a genefically-idenfical populafion of cells will be investigated. An understanding of the resulfing phenotypic diversity of tumor cells will lead to improved therapeufic intervenfions eradicafing tumor populafions. Specifically, we will (1) characterize the variable response of tumor cells to growth factors and targeted inhibitors: (2) design a mathemafical framework to predict the consequences of cellular diversity and identify the opfimum therapeutic intervenfion that maximizes the chance of eradicafing the tumor;and (3) validate the predictions of the mathemafical framework in cell lines and murine models. This project fully leverages our expertise in cancer biology and clinical experience (Bromberg), single cell profiling and biochemical modeling (Altan-Bonnet), and mathemafical modeling (Michor). We will dissect how the IL-6 pathway can generate phenotypic variability in tumors, which drives their progression and causes resistance to targeted therapies [7- 13]. Based on our models, we will identify and test the opfimal therapeufic protocol for treafing these cancers.